7 |
Catalyst::Model |
Catalyst::Model |
8 |
/; |
/; |
9 |
use Data::Dumper; |
use Data::Dumper; |
10 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.08; |
11 |
use WebPAC::Output::TT 0.02; |
use WebPAC::Output::TT 0.07; |
12 |
use WebPAC::Search::Estraier 0.05; |
use WebPAC::Search::Estraier 0.05; |
13 |
use File::Slurp; |
use File::Slurp; |
14 |
use Time::HiRes; |
use Time::HiRes; |
15 |
|
use Encode qw/encode decode from_to/; |
16 |
|
use Data::HTMLDumper; |
17 |
|
|
18 |
=head1 NAME |
=head1 NAME |
19 |
|
|
42 |
user: 'admin' |
user: 'admin' |
43 |
passwd: 'admin' |
passwd: 'admin' |
44 |
hits_on_page: 100 |
hits_on_page: 100 |
45 |
|
hits_for_pager: 1000 |
46 |
|
|
47 |
webpac: |
webpac: |
48 |
db_path: '/data/webpac2/db' |
db_path: '/data/webpac2/db' |
50 |
template: 'html_ffzg_results_short.tt' |
template: 'html_ffzg_results_short.tt' |
51 |
# encoding comming from webpac |
# encoding comming from webpac |
52 |
webpac_encoding: 'iso-8859-2' |
webpac_encoding: 'iso-8859-2' |
|
# encoding expected by Catalyst |
|
|
out_encoding: 'UTF-8' |
|
53 |
|
|
54 |
=cut |
=cut |
55 |
|
|
65 |
my $est_cfg = $c->config->{hyperestraier}; |
my $est_cfg = $c->config->{hyperestraier}; |
66 |
$est_cfg->{'log'} = $log; |
$est_cfg->{'log'} = $log; |
67 |
|
|
68 |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding}; |
$est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding"); |
69 |
|
|
70 |
$log->debug("using config:" . Dumper($est_cfg) ); |
$log->debug("using config:" . Dumper($est_cfg) ); |
71 |
|
|
78 |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } ); |
79 |
|
|
80 |
# save config parametars in object |
# save config parametars in object |
81 |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding out_encoding defaultdepth/) { |
foreach my $f (qw/db_path template_path hits_on_page webpac_encoding defaultdepth/) { |
82 |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
$self->{$f} = $c->config->{hyperestraier}->{$f} || |
83 |
$c->config->{webpac}->{$f}; |
$c->config->{webpac}->{$f}; |
84 |
$log->debug("self->{$f} = " . $self->{$f}); |
$log->debug("self->{$f} = " . $self->{$f}); |
96 |
|
|
97 |
$self->{out} = new WebPAC::Output::TT( |
$self->{out} = new WebPAC::Output::TT( |
98 |
include_path => $template_path, |
include_path => $template_path, |
99 |
filters => { foo => sub { shift } }, |
filters => { |
100 |
|
dump_html => sub { |
101 |
|
my $t = shift || return; |
102 |
|
#return Data::HTMLDumper->Dumper( $t ); |
103 |
|
return Data::HTMLDumper->Dump([$t],[qw/dump/]); |
104 |
|
} |
105 |
|
}, |
106 |
); |
); |
107 |
|
|
108 |
# default template from config.yaml |
# default template from config.yaml |
109 |
$self->{template} ||= $c->config->{webpac}->{template}; |
$self->{template} ||= $c->config->{webpac}->{template}; |
110 |
|
|
|
$self->{iconv} = new Text::Iconv( |
|
|
$c->config->{webpac}->{webpac_encoding}, |
|
|
$c->config->{webpac}->{out_encoding} |
|
|
); |
|
|
|
|
111 |
$log->debug("converting encoding from webpac_encoding '" . |
$log->debug("converting encoding from webpac_encoding '" . |
112 |
$c->config->{webpac}->{webpac_encoding} . |
$c->config->{webpac}->{webpac_encoding} . |
|
"' to '" . |
|
|
$c->config->{webpac}->{out_encoding} . |
|
113 |
"'" |
"'" |
114 |
); |
); |
115 |
|
|
116 |
|
$self->{databases} = $c->config->{databases} || $log->error("can't find databases in config"); |
117 |
|
|
118 |
return $self; |
return $self; |
119 |
|
|
144 |
|
|
145 |
my $log = $self->{log}; |
my $log = $self->{log}; |
146 |
|
|
147 |
$log->debug("args: " . Dumper( $args )); |
$log->debug("search args: " . Dumper( $args )); |
148 |
|
|
149 |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
my $query = $args->{phrase} || $log->warn("no query phrase") && return; |
150 |
|
|
157 |
|
|
158 |
my $template_filename = $args->{template} || $self->{template}; |
my $template_filename = $args->{template} || $self->{template}; |
159 |
|
|
160 |
$args->{max} ||= $self->{'hits_on_page'}; |
$args->{max} ||= $self->{'hits_for_pager'}; |
161 |
if (! $args->{max}) { |
if (! $args->{max}) { |
162 |
$args->{max} = 10; |
$args->{max} = 100; |
163 |
$log->warn("max not set when calling model. Using default of 10"); |
$log->warn("max not set when calling model. Using default of $args->{max}"); |
164 |
} |
} |
165 |
|
|
166 |
my $times; # store some times for benchmarking |
my $times; # store some times for benchmarking |
193 |
|
|
194 |
for my $i ( 0 .. $#results ) { |
for my $i ( 0 .. $#results ) { |
195 |
|
|
196 |
my ($id, $prefix); |
my ($database, $prefix, $id); |
197 |
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)#(\d+)$!) { |
if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) { |
198 |
($prefix,$id) = ($1,$2); |
($database, $prefix,$id) = ($1,$2,$3); |
199 |
} else { |
} else { |
200 |
$log->warn("can't decode prefix#id from " . $results[$i]->{'@uri'}); |
$log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'}); |
201 |
next; |
next; |
202 |
} |
} |
203 |
|
|
205 |
|
|
206 |
$t = time(); |
$t = time(); |
207 |
|
|
208 |
my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix ) || $log->error("can't load_ds( id => $id, prefix => '$prefix' )") && next; |
my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id ); |
209 |
|
if (! $ds) { |
210 |
|
$log->error("can't load_ds( ${database}/${prefix}/${id} )"); |
211 |
|
next; |
212 |
|
} |
213 |
|
|
214 |
$times->{db} += time() - $t; |
$times->{db} += time() - $t; |
215 |
|
|
220 |
my $html = $self->{out}->apply( |
my $html = $self->{out}->apply( |
221 |
template => $template_filename, |
template => $template_filename, |
222 |
data => $ds, |
data => $ds, |
223 |
|
record_uri => "${database}/${prefix}/${id}", |
224 |
|
config => $self->{databases}->{$database}, |
225 |
); |
); |
226 |
|
|
227 |
$times->{out} += time() - $t; |
$times->{out} += time() - $t; |
228 |
|
|
229 |
$t = time(); |
$t = time(); |
230 |
|
|
231 |
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
$html = decode($self->{webpac_encoding}, $html); |
|
|
|
|
$times->{iconv} += time() - $t; |
|
232 |
|
|
233 |
push @html_results, $html; |
push @html_results, $html; |
234 |
|
|
237 |
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
#$log->debug( '@html_results = ' . Dumper( \@html_results ) ); |
238 |
|
|
239 |
$log->debug( sprintf( |
$log->debug( sprintf( |
240 |
"time spent: db = %.2f, out = %.2f, iconv = %.2f", |
"time spent: db = %.2f, out = %.2f", |
241 |
$times->{db}, $times->{out}, $times->{iconv}, |
$times->{db}, $times->{out}, |
242 |
) ); |
) ); |
243 |
|
|
244 |
return \@html_results; |
return \@html_results; |
261 |
|
|
262 |
my $args = {@_}; |
my $args = {@_}; |
263 |
my $log = $self->{log}; |
my $log = $self->{log}; |
264 |
$log->debug("args: " . Dumper( $args )); |
$log->debug("record args: " . Dumper( $args )); |
265 |
|
|
266 |
foreach my $f (qw/mfn template prefix/) { |
foreach my $f (qw/record_uri template/) { |
267 |
$log->die("need $f") unless ($args->{$f}); |
$log->fatal("need $f") unless ($args->{$f}); |
268 |
} |
} |
269 |
|
|
270 |
my $mfn = $args->{mfn}; |
my ($database, $prefix, $id); |
271 |
|
|
272 |
|
if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) { |
273 |
|
($database, $prefix, $id) = ($1,$2,$3); |
274 |
|
} else { |
275 |
|
$log->error("can't parse $args->{record_uri} into prefix, database and uri"); |
276 |
|
return; |
277 |
|
} |
278 |
|
|
279 |
my $ds = $self->{db}->load_ds( id => $mfn, prefix => 'FIXME' ) || $log->error("can't load_ds( $mfn )") && next; |
my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database ); |
280 |
|
if (! $ds) { |
281 |
|
$log->error("can't load_ds( $database/$prefix/$id )"); |
282 |
|
return; |
283 |
|
} |
284 |
|
|
285 |
my $html = $self->{out}->apply( |
my $html = $self->{out}->apply( |
286 |
template => $args->{template}, |
template => $args->{template}, |
287 |
data => $ds, |
data => $ds, |
288 |
|
record_uri => $args->{record_uri}, |
289 |
|
config => $self->{databases}->{$database}, |
290 |
); |
); |
291 |
|
|
292 |
$html = $self->{iconv}->convert( $html ) || $log->error("can't convert: $html"); |
$html = decode($self->{webpac_encoding}, $html); |
293 |
|
|
294 |
return $html; |
return $html; |
295 |
} |
} |
296 |
|
|
297 |
|
|
298 |
=head2 save_html |
=head2 save_html |
299 |
|
|
300 |
$m->save_html( '/full/path/to/file', $content ); |
$m->save_html( '/full/path/to/file', $content ); |
301 |
|
|
302 |
It will use C<Text::Iconv> to convert content encoding back to |
It will use C<Encode> to convert content encoding back to |
303 |
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
Webpac codepage, recode JavaScript Unicode entities (%u1234), |
304 |
strip extra newlines at beginning and end, and save to |
strip extra newlines at beginning and end, and save to |
305 |
C</full/path/to/file.new> and if that succeeds, just rename |
C</full/path/to/file.new> and if that succeeds, just rename |
310 |
sub save_html { |
sub save_html { |
311 |
my ($self, $path, $content) = @_; |
my ($self, $path, $content) = @_; |
312 |
|
|
313 |
|
# FIXME Should this be UTF-8 or someting? |
314 |
|
my $js_encoding = $self->{webpac_encoding}; |
315 |
|
$js_encoding = 'UTF-16'; |
316 |
|
|
317 |
sub _conv_js { |
sub _conv_js { |
318 |
my $t = shift || return; |
return '0x' . $_[1]; |
319 |
return $self->{iconv}->convert(chr(hex($t))); |
return encode($_[0], chr(hex($_[1]))); |
320 |
} |
} |
321 |
$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($1)/gex; |
#$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex; |
322 |
$content =~ s/^[\n\r]+//s; |
$content =~ s/^[\n\r]+//s; |
323 |
$content =~ s/[\n\r]+$/\n/s; |
$content =~ s/[\n\r]+$/\n/s; |
324 |
|
$content =~ s/\n\r/\n/gs; |
325 |
|
|
326 |
my ($from, $to) = ( |
my $disk_encoding = $self->{webpac_encoding} || 'utf-8'; |
327 |
$self->{out_encoding}, |
$self->{log}->debug("convert encoding to $disk_encoding"); |
328 |
$self->{webpac_encoding}, |
from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content"); |
|
); |
|
|
|
|
|
$self->{log}->debug("using iconv to convert from $from to $to encoding"); |
|
|
|
|
|
my $iconv_on_save = new Text::Iconv($from, $to) |
|
|
|| $self->{log}->fatal("can't create iconv for saving"); |
|
|
|
|
|
$content = $iconv_on_save->convert( $content ) || die "no content?"; |
|
329 |
|
|
330 |
write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!"; |
write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!"; |
331 |
rename $path . '.new', $path || die "can't rename to $path: $!"; |
rename $path . '.new', $path || die "can't rename to $path: $!"; |
347 |
die "no path?" unless ($path); |
die "no path?" unless ($path); |
348 |
|
|
349 |
my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!"; |
my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!"; |
|
#$content = $q->escapeHTML($iconv_utf8->convert($content)); |
|
|
$content = $self->{iconv}->convert($content); |
|
350 |
|
|
351 |
return $content; |
return decode($self->{webpac_encoding}, $content); |
352 |
} |
} |
353 |
|
|
354 |
=head1 AUTHOR |
=head1 AUTHOR |
355 |
|
|
356 |
Dobrica Pavlinusic |
Dobrica Pavlinusic C<< <dpavlin@rot13.org> >> |
357 |
|
|
358 |
=head1 LICENSE |
=head1 LICENSE |
359 |
|
|